[Printing-architecture] Printer setup tools: Automatically choose the correct backend

[Printing-architecture] Printer setup tools: Automatically choose the correct backend

[Till Kamppeter]

Hi,

currently, it is no problem any more to determine which model a 
connected printer is and to find the correct PPD/driver for it. The 
bigger problem is to use the correct backend automatically if the driver 
comes with its own backend. Examples for such drivers are Gutenprint and 
HPLIP. The "canon" and "epson" backends for Gutenprint are usually not 
used and this does not restrict anything on normal printing. They only 
allow printer maintenance "through the print queue" by sending special 
commands. But for HPLIP the "hp" and "hpfax" backends are very 
important. The first allows printing with bi-directional functionality 
(HPLIP needs this also for border-less printing) and the second allows 
faxing on HP's multi-function devices.

The list of auto-detected printers in a printer setup tool is the 
equivalent of the output of "lpinfo -v":

------------------------------------------------------------------------
till@till-laptop:~/printing/foomatic/foomatic-db-hpijs$ lpinfo -v
network socket
network beh
direct 
hal:///org/freedesktop/Hal/devices/usb_device_3f0_3517_00CNMJP81545_if0_printer_noserial
network socket://192.168.2.34
network ipp://fe80::21d:60ff:fe48:c2d7:631/printers/SanFranciscoFax
network socket://192.168.2.29
direct hpfax:/usb/HP_LaserJet_3390?serial=00CNMJP81545
direct usb://HP/LaserJet%203390
direct hp:/usb/HP_LaserJet_3390?serial=00CNMJP81545
direct 
hal:///org/freedesktop/Hal/devices/usb_device_3f0_2812_CN7BU18154_if0_printer_CN7BU18154
direct usb://HP/Officejet%20H470?serial=CN7BU18154
direct hp:/usb/Officejet_H470?serial=CN7BU18154
network socket://192.168.2.35
direct 
hal:///org/freedesktop/Hal/devices/usb_device_3f0_bd02_MY7143104604C1_if0_printer_MY7143104604C1
network socket://192.168.2.30
direct usb://HP/Photosmart%20Pro%20B9100%20series?serial=MY7143104604C1
direct hp:/usb/Photosmart_Pro_B9100_series?serial=MY7143104604C1
network http
network ipp
network lpd
network bluetooth://001A0E1769AA
file cups-pdf:/
direct scsi
network smb
till@till-laptop:~/printing/foomatic/foomatic-db-hpijs$
------------------------------------------------------------------------

You see already that several printers get detected three times, first by 
the "hal" backend, second by "usb" and third by "hp". Users of a GUI- or 
web-based printer setup tool will see the model name of their printer 
three times and are completely confused what they should choose. So only 
a few will choose the "hp" backend and so will be able to print 
borderless or use the hp-toolbox.

A method which I have applied twice (printerdrake and 
system-config-printer) is to add explicit support for backends supplied 
with the driver. Therefore at least these two backends select the "hp" 
backend automatically (and even suppress the superfluous entries in the 
list of detected printers). This works fine at least for drivers which 
are common enough and existed before the release of the printer setup 
tool in use. Some authors also give no special support for third-party 
software in general.

So one needs a method to auto-select the best backend for a printer. As 
all information about each supported printer model is supplied to CUPS 
by a PPD file for the printer, the most obvious place for information 
about the preferred backend is the PPD file.

Therefore I suggest the following PPD extensions:

*cupsPreferBackend: "hp"

This entry simply recommends a backend. In this case a printer setup 
tool should look for full URIs of this backend in the output of "lpinfo 
-l -v" and compares make-and-model and device ID entries with the ones 
of the URI which the user has chosen. If both the user-chosen URI qnd 
the URI found as replacement have the printer's serial number or IP, 
they must match. URIs with serial number or IP should never be replaced 
by URIs with another serial number or IP.

*cupsPreferBackend usb,parallel,hal: "hp"

This is a modifier for all keywords listed here. It estricts the 
application of the backend selection rule to the case when the URI 
selected by the user uses one of the listed backends, in this example 
the printer setup tool should only try to use the "hp" backend if the 
user has selected an URI for the "usb", "parallel", or "hal" backend.

*cupsReplaceURI: "usb://HP/LaserJet(_|%20)42\d\d hp:/usb/HP_LaserJet_42\d\d"

Here we do not use model info from the "lpinfo -l -v" output. The 
printer setup tool should match the first regexp against the user-chosen 
URI and the search the "lpinfo -v" output for an URI matching the second 
regexp. In this case the user-chosen URI gets relaced. Serial numbers 
and IPs have to get checked here, too, as with the *cupsPreferBackend 
keyword.

*cupsMakeURI: "hp-makeuri -c @IP@"
*cupsMakeURI: "hp-makeuri -c @USB_BUS@:@USB_DEV@"
*cupsMakeURI: "hp-makeuri -c @PARDEV@"

*cupsMakeURI: "foojet-geturi @USBDEV@"
*cupsMakeURI: "foojet-geturi @URI@"

Some drivers have their own tools to determine the correct URIs for 
their printers. With these lines a printer setup tool should execute the 
shown commands, substituting the parameters (@...@) at first. There can 
be more than one line, and they should be tried from the top to the 
bottom, until one line produces a valid URI. So if a parameter cannot be 
substituted because it does not exist, as for example the IP for an USB 
printer, or if a command gives an error, the next line gets tried. If 
all lines fail the printer setup tool creates the queue, but it uses the 
URI chosen by the user.

@URI@ is here the URI chosen by the user, @PARDEV@ is the device file 
for the parallel port, like /dev/lp0, @USBDEV@ is the USB device like 
/dev/usb/lp0, ...

In general there should be more than one of all the above keywords 
allowed and substitutions be attempted using each line, from the top to 
the bottom, until the first line appears which generates a valid 
replacement URI.

*cupsMakeExtraURI lsb/usr/hpijs/HP/HP-Fax-hplip.ppd: "hp-makeuri -f @IP@"

This keyword is for multi-function devices. It calls a tool to find out 
URIs for additional queues, like for example a fax queue. A printer 
setup tool is supposed to finish the setup of the current print queue 
and then to create a queue for the URI found here. The CUPS URI for the 
PPD file for that extra queue is given before the colon.

There can be more than one of these lines. In this case for each 
succeeding line (except duplicate URIs) one extra queue has to be created.

*cupsExtraURI lsb/usr/hpijs/HP/HP-Fax-hplip.ppd: "s!^hp:!hpfax:!"

On the resulting URI for this print queue (after substitution applying 
the previous keywords) the given regexp substitution is applied. If the 
substitution succeeds, the printer setup tool is supposed to create an 
additional queue with the given PPD file.

What do you think about this PPD extension? This should finally allow to 
set up printers completely automatically and non-interactively. This 
should allow Plug'n'Print in all cases and also with multi-function devices.

    Till

Re: [Printing-architecture] Printer setup tools: Automatically choose the correct backend

[Till Kamppeter]

Mike Sweet does not like this idea. He wrote on the cups.development 
newsgroup:

---------------------------------------------------------------------------
I think it would lead to an unending stream of compatibility and
other problems.

There is no reason for having multiple backends that do the same
thing.  HP's backend is unnecessary once we use libusb instead of
the Linux USB printer character device, and that is going to happen
in CUPS 1.4 (and if I understood correctly is already used by several
Linux distros).  For network printer, HP should be using the socket
backend.  The canon and epson backends have been removed from
Gutenprint.  There should be no "hal" backend, either.

In short, once you get rid of the extra unnecessary backends, there
is no problem selecting the "right" backend for a PPD.  Futhermore,
a fundamental design goal of CUPS is that a given CUPS driver MUST
work with ANY backend.  Drivers that do not work with all backends
are by definition broken.
---------------------------------------------------------------------------

Here I would like to hear especially the opinion of the printer 
manufacturers. Are custom backends really needed? Or would it be 
possible to make all drivers work with the backends which come with CUPS?

    Till


Till Kamppeter wrote:
> Hi,
> 
> currently, it is no problem any more to determine which model a 
> connected printer is and to find the correct PPD/driver for it. The 
> bigger problem is to use the correct backend automatically if the driver 
> comes with its own backend. Examples for such drivers are Gutenprint and 
> HPLIP. The "canon" and "epson" backends for Gutenprint are usually not 
> used and this does not restrict anything on normal printing. They only 
> allow printer maintenance "through the print queue" by sending special 
> commands. But for HPLIP the "hp" and "hpfax" backends are very 
> important. The first allows printing with bi-directional functionality 
> (HPLIP needs this also for border-less printing) and the second allows 
> faxing on HP's multi-function devices.
> 
> The list of auto-detected printers in a printer setup tool is the 
> equivalent of the output of "lpinfo -v":
> 
> ------------------------------------------------------------------------
> till@till-laptop:~/printing/foomatic/foomatic-db-hpijs$ lpinfo -v
> network socket
> network beh
> direct 
> hal:///org/freedesktop/Hal/devices/usb_device_3f0_3517_00CNMJP81545_if0_printer_noserial 
> 
> network socket://192.168.2.34
> network ipp://fe80::21d:60ff:fe48:c2d7:631/printers/SanFranciscoFax
> network socket://192.168.2.29
> direct hpfax:/usb/HP_LaserJet_3390?serial=00CNMJP81545
> direct usb://HP/LaserJet%203390
> direct hp:/usb/HP_LaserJet_3390?serial=00CNMJP81545
> direct 
> hal:///org/freedesktop/Hal/devices/usb_device_3f0_2812_CN7BU18154_if0_printer_CN7BU18154 
> 
> direct usb://HP/Officejet%20H470?serial=CN7BU18154
> direct hp:/usb/Officejet_H470?serial=CN7BU18154
> network socket://192.168.2.35
> direct 
> hal:///org/freedesktop/Hal/devices/usb_device_3f0_bd02_MY7143104604C1_if0_printer_MY7143104604C1 
> 
> network socket://192.168.2.30
> direct usb://HP/Photosmart%20Pro%20B9100%20series?serial=MY7143104604C1
> direct hp:/usb/Photosmart_Pro_B9100_series?serial=MY7143104604C1
> network http
> network ipp
> network lpd
> network bluetooth://001A0E1769AA
> file cups-pdf:/
> direct scsi
> network smb
> till@till-laptop:~/printing/foomatic/foomatic-db-hpijs$
> ------------------------------------------------------------------------
> 
> You see already that several printers get detected three times, first by 
> the "hal" backend, second by "usb" and third by "hp". Users of a GUI- or 
> web-based printer setup tool will see the model name of their printer 
> three times and are completely confused what they should choose. So only 
> a few will choose the "hp" backend and so will be able to print 
> borderless or use the hp-toolbox.
> 
> A method which I have applied twice (printerdrake and 
> system-config-printer) is to add explicit support for backends supplied 
> with the driver. Therefore at least these two backends select the "hp" 
> backend automatically (and even suppress the superfluous entries in the 
> list of detected printers). This works fine at least for drivers which 
> are common enough and existed before the release of the printer setup 
> tool in use. Some authors also give no special support for third-party 
> software in general.
> 
> So one needs a method to auto-select the best backend for a printer. As 
> all information about each supported printer model is supplied to CUPS 
> by a PPD file for the printer, the most obvious place for information 
> about the preferred backend is the PPD file.
> 
> Therefore I suggest the following PPD extensions:
> 
> *cupsPreferBackend: "hp"
> 
> This entry simply recommends a backend. In this case a printer setup 
> tool should look for full URIs of this backend in the output of "lpinfo 
> -l -v" and compares make-and-model and device ID entries with the ones 
> of the URI which the user has chosen. If both the user-chosen URI qnd 
> the URI found as replacement have the printer's serial number or IP, 
> they must match. URIs with serial number or IP should never be replaced 
> by URIs with another serial number or IP.
> 
> *cupsPreferBackend usb,parallel,hal: "hp"
> 
> This is a modifier for all keywords listed here. It estricts the 
> application of the backend selection rule to the case when the URI 
> selected by the user uses one of the listed backends, in this example 
> the printer setup tool should only try to use the "hp" backend if the 
> user has selected an URI for the "usb", "parallel", or "hal" backend.
> 
> *cupsReplaceURI: "usb://HP/LaserJet(_|%20)42\d\d 
> hp:/usb/HP_LaserJet_42\d\d"
> 
> Here we do not use model info from the "lpinfo -l -v" output. The 
> printer setup tool should match the first regexp against the user-chosen 
> URI and the search the "lpinfo -v" output for an URI matching the second 
> regexp. In this case the user-chosen URI gets relaced. Serial numbers 
> and IPs have to get checked here, too, as with the *cupsPreferBackend 
> keyword.
> 
> *cupsMakeURI: "hp-makeuri -c @IP@"
> *cupsMakeURI: "hp-makeuri -c @USB_BUS@:@USB_DEV@"
> *cupsMakeURI: "hp-makeuri -c @PARDEV@"
> 
> *cupsMakeURI: "foojet-geturi @USBDEV@"
> *cupsMakeURI: "foojet-geturi @URI@"
> 
> Some drivers have their own tools to determine the correct URIs for 
> their printers. With these lines a printer setup tool should execute the 
> shown commands, substituting the parameters (@...@) at first. There can 
> be more than one line, and they should be tried from the top to the 
> bottom, until one line produces a valid URI. So if a parameter cannot be 
> substituted because it does not exist, as for example the IP for an USB 
> printer, or if a command gives an error, the next line gets tried. If 
> all lines fail the printer setup tool creates the queue, but it uses the 
> URI chosen by the user.
> 
> @URI@ is here the URI chosen by the user, @PARDEV@ is the device file 
> for the parallel port, like /dev/lp0, @USBDEV@ is the USB device like 
> /dev/usb/lp0, ...
> 
> In general there should be more than one of all the above keywords 
> allowed and substitutions be attempted using each line, from the top to 
> the bottom, until the first line appears which generates a valid 
> replacement URI.
> 
> *cupsMakeExtraURI lsb/usr/hpijs/HP/HP-Fax-hplip.ppd: "hp-makeuri -f @IP@"
> 
> This keyword is for multi-function devices. It calls a tool to find out 
> URIs for additional queues, like for example a fax queue. A printer 
> setup tool is supposed to finish the setup of the current print queue 
> and then to create a queue for the URI found here. The CUPS URI for the 
> PPD file for that extra queue is given before the colon.
> 
> There can be more than one of these lines. In this case for each 
> succeeding line (except duplicate URIs) one extra queue has to be created.
> 
> *cupsExtraURI lsb/usr/hpijs/HP/HP-Fax-hplip.ppd: "s!^hp:!hpfax:!"
> 
> On the resulting URI for this print queue (after substitution applying 
> the previous keywords) the given regexp substitution is applied. If the 
> substitution succeeds, the printer setup tool is supposed to create an 
> additional queue with the given PPD file.
> 
> What do you think about this PPD extension? This should finally allow to 
> set up printers completely automatically and non-interactively. This 
> should allow Plug'n'Print in all cases and also with multi-function 
> devices.
> 
>    Till
> 

Re: [Printing-architecture] Printer setup tools: Automatically choose the correct backend

[Till Kamppeter]

Suffield, David wrote:
> Hi,
> 
> I agree there are too many backends for the same printer. In our case I would like to see only one backend selection for HP printers.
>

It would be great to have only one backend per connection type, but it 
seems that printers have a lot of functionality which use 
manufacturer-specific low-level protocols and for that the manufacturers 
want to create their own backends. Manufacturers also work around 
hardware/firmware bugs by the software, which could also lead to a new 
backend.

> I would consider using the Mike's "usb" backend if it meets our needs. We need a backend that provides printing and status over various types of IO (ie: raw, 1284.4 and MLC). The IO can be very manufacturer and product specific and I don't see how the current "usb" backend can provide this functionality.

Here I can imagine that Mike Sweet has no problems to provide raw and 
1284.4 access with his backend, but AFAIK MLC is HP-specific and as far 
as I know Mike, he would not implement that.

For me this situation would mean that we need backend hinting in the PPDs.

> 
> I have looked at cups-1.4svn-r7679-source.tar.gz and I don't see the libusb support in the "usb" backend am I missing something?
> 
> Even if we cannot use Mike's "usb" backend I would greatly appreciate the libusb support. Right now hal_lpadmin is a pain because it does not understand backends that do not use usblp.
>

This is not hal_lpadmin (I am one of the upstream developers), it is HAL 
itself. HAL gives printer add and remove signals depending on signals 
from the usblp kernel module, it will not recognize printers at all if 
one would move the module away. HAL needs a printer recognition which 
works without the usblp module.

> In FC9 hal_lpadmin can turn a print queue off when usblp is removed because it assumes the printer is off line. The user has to manually restart the queue in order to print.
>

As a workaround I have excluded HP printers from being disabled by 
hal_lpadmin when they are turned off.

    Till

Re: [Printing-architecture] [Gimp-print-devel] Printer setup tools: Automatically choose the correct backend

[Till Kamppeter]

Robert Krawitz wrote:
> 
> This problem's not restricted to HP printers -- I have the same
> problem with Epson printers.  If I turn the printer off, the queue
> gets disabled, and then I have to manually re-enable the queue when I
> turn the printer back on.  This is with OpenSUSE 10.3.
> 

The queues are supposed to get turned back on when reconnecting the printer.

    Till

Re: [Printing-architecture] [Gimp-print-devel] Printer setup tools: Automatically choose the correct backend

[Klaus Singvogel]

Till Kamppeter wrote:
> Robert Krawitz wrote:
> > 
> > This problem's not restricted to HP printers -- I have the same
> > problem with Epson printers.  If I turn the printer off, the queue
> > gets disabled, and then I have to manually re-enable the queue when I
> > turn the printer back on.  This is with OpenSUSE 10.3.
> > 
> 
> The queues are supposed to get turned back on when reconnecting the printer.


I know that we got a bugzilla on this topic for 10.3.
But I'm sorry, I don't know, if it has been dealt. I'm not the author
nor the responsible manager here.

But you can get the information, if you search at our bugzilla for
this entry:
        https://bugzilla.novell.com

I can tell you, that we released a bugfix for "cups-autoconf" in the
past for 10.3. - Maybe you just want to update your system?

Regards,
	Klaus.
-- 
Klaus Singvogel  -  Maxfeldstr. 5  -  90409 Nuernberg  -  Germany
Phone: +49-911-74053-0
GnuPG-Key-ID: 1024R/5068792D  1994-06-27
SUSE LINUX Products GmbH,  GF: Markus Rex,  HRB 16746 (AG Nuernberg)

Re: [Printing-architecture] [Gimp-print-devel] Printer setup tools: Automatically choose the correct backend

[Till Kamppeter]

Does SUSE really use the hal-cups-utils? They require 
system-config-printer and AFAIK this is not available on SUSE.

    Till

Robert Krawitz wrote:
>    From: Tim Waugh <twaugh@redhat.com>
>    Date: Mon, 23 Jun 2008 12:02:53 +0100
> 
>    On Mon, 2008-06-23 at 06:52 -0400, Robert Krawitz wrote:
>    > This problem's not restricted to HP printers -- I have the same
>    > problem with Epson printers.  If I turn the printer off, the queue
>    > gets disabled, and then I have to manually re-enable the queue when I
>    > turn the printer back on.  This is with OpenSUSE 10.3.
> 
>    Which version of hal-cups-utils?
> 
> This is all I can find.  It may have been an unwarranted guess on my
> part that hal is responsible for this, but the effect is still there.
> 
> [1(rlk)|{!55}<dsl092-065-009>/home/rlk/sandbox/print-5.1/src/testpattern]
> $ rpm -qa |grep cups
> libgnomecups-lang-0.2.3-53.1
> cups-devel-1.2.12-22.15
> gnome-cups-manager-0.32cvs20060120-169.2
> libgnomecups-devel-0.2.3-53.1
> cups-client-1.2.12-22.15
> gnome-cups-manager-devel-0.32cvs20060120-169.2
> cups-libs-1.2.12-22.15
> cupsddk-1.2.0-27
> cups-autoconfig-0.1.0-27.2
> rpmkey-apcupsd-0.1-2
> cups-backends-1.0-126
> libgnomecups-0.2.3-53.1
> cups-drivers-1.2.12-28
> cups-1.2.12-22.15
> [1(rlk)|{!56}<dsl092-065-009>/home/rlk/sandbox/print-5.1/src/testpattern]
> $ rpm -qa |grep hal
> hal-palm-0.12.2-64.20
> phalanx-22-583
> hal-0.5.9_git20070831-13.2
> hal-devel-0.5.9_git20070831-13.2
> hal-resmgr-0.1_SVNr141-29
> 

Re: [Printing-architecture] [Gimp-print-devel] Printer setup tools: Automatically choose the correct backend

[Klaus Singvogel]

Hello,
SUSE is using "cups-autoconfig" for automatic configuration of
printers when a plug-in event in hal occurs. It's not
"hal-cups-utils", it's a new tool, which got developed in-house.

I didn't get the full thread on this mailing-list, so I don't
know, if already answered. So I hope nobody minds if I repeat
the same answer.

Regards,
	Klaus.

Till Kamppeter wrote:
> Does SUSE really use the hal-cups-utils? They require 
> system-config-printer and AFAIK this is not available on SUSE.
> 
>     Till
> 
> Robert Krawitz wrote:
> >    From: Tim Waugh <twaugh@redhat.com>
> >    Date: Mon, 23 Jun 2008 12:02:53 +0100
> > 
> >    On Mon, 2008-06-23 at 06:52 -0400, Robert Krawitz wrote:
> >    > This problem's not restricted to HP printers -- I have the same
> >    > problem with Epson printers.  If I turn the printer off, the queue
> >    > gets disabled, and then I have to manually re-enable the queue when I
> >    > turn the printer back on.  This is with OpenSUSE 10.3.
> > 
> >    Which version of hal-cups-utils?
> > 
> > This is all I can find.  It may have been an unwarranted guess on my
> > part that hal is responsible for this, but the effect is still there.
> > 
> > [1(rlk)|{!55}<dsl092-065-009>/home/rlk/sandbox/print-5.1/src/testpattern]
> > $ rpm -qa |grep cups
> > libgnomecups-lang-0.2.3-53.1
> > cups-devel-1.2.12-22.15
> > gnome-cups-manager-0.32cvs20060120-169.2
> > libgnomecups-devel-0.2.3-53.1
> > cups-client-1.2.12-22.15
> > gnome-cups-manager-devel-0.32cvs20060120-169.2
> > cups-libs-1.2.12-22.15
> > cupsddk-1.2.0-27
> > cups-autoconfig-0.1.0-27.2
> > rpmkey-apcupsd-0.1-2
> > cups-backends-1.0-126
> > libgnomecups-0.2.3-53.1
> > cups-drivers-1.2.12-28
> > cups-1.2.12-22.15
> > [1(rlk)|{!56}<dsl092-065-009>/home/rlk/sandbox/print-5.1/src/testpattern]
> > $ rpm -qa |grep hal
> > hal-palm-0.12.2-64.20
> > phalanx-22-583
> > hal-0.5.9_git20070831-13.2
> > hal-devel-0.5.9_git20070831-13.2
> > hal-resmgr-0.1_SVNr141-29
> > 
> 
> _______________________________________________
> Printing-architecture mailing list
> Printing-architecture@lists.linux-foundation.org
> https://lists.linux-foundation.org/mailman/listinfo/printing-architecture

-- 
Klaus Singvogel  -  Maxfeldstr. 5  -  90409 Nuernberg  -  Germany
Phone: +49-911-74053-0
GnuPG-Key-ID: 1024R/5068792D  1994-06-27
SUSE LINUX Products GmbH,  GF: Markus Rex,  HRB 16746 (AG Nuernberg)

Re: [Printing-architecture] [Gimp-print-devel] Printer setup tools: Automatically choose the correct backend

[Till Kamppeter]

David Rosky wrote:
> 
> 
> 
>     Why does HPLIP need to talk to the printer to do borderless printing?
>     Is it to verify the paper size that's actually in the printer?  If so,
>     you could either trust the paper size the user specifies.
> 
> 
> Some drivers don't want to allow borderless printing if a certain *type* 
> of paper isn't loaded (e.g., photo paper).  Perhaps HPLIP is checking 
> that.  Personally, however, I'd be in favor of trusting the user, as has 
> generally been the gutenprint philosophy.
> 

There are some printers which do borderless only for 5x7-inch-photos and 
smaller and some do no borderless at all, but all the rest of the 
communication protocol of these printers is the same as for the 
full-borderless models. For me it seems that the driver asks the printer 
which borderless type it is (and perhaps even also the loaded paper 
size) to not make the printer with a borderless job sent when the 
printer is not capable to print it borderless. Am I correct, Dave?

    Till

Re: [Printing-architecture] [Gimp-print-devel] Printer setup tools: Automatically choose the correct backend

[Till Kamppeter]

Robert Krawitz wrote:
> That implies that the driver is trying to double check what the user
> specified -- the user already has to specify the paper size and media
> type.  Things shouldn't fail if the driver's not able to verify this.
>

Dave, Shiyun, as you are using CUPS DDK now to generate PPDs, you should 
introduce option conflicts/UI constraints in the PPDs to avoid 
borderless printing on unsuitable paper sizes.

    Till

Re: [Printing-architecture] [Gimp-print-devel] Printer setup tools: Automatically choose the correct backend

[Till Kamppeter]

Michael R Sweet wrote:
> Suffield, David wrote:
>> ...
>> I would consider using the Mike's "usb" backend if it meets our
>> needs. We need a backend that provides printing and status over
>> various types of IO (ie: raw, 1284.4 and MLC). The IO can be very
>> manufacturer and product specific and I don't see how the current
>> "usb" backend can provide this functionality.
> 
> Our current focus is on supporting libusb; if necessary, we can add
> a side-channel request to set the IO mode, but otherwise I suspect
> we can use the default mode supported by the device.
> 
> We already support a variety of side-channel data - 1284 device ID,
> centronics bits, bidirectional support, SNMP OIDs (for network
> printers), etc.
>

Will this really cover all manufacturer-specific access methods?

>> I have looked at cups-1.4svn-r7679-source.tar.gz and I don't see the
>> libusb support in the "usb" backend am I missing something?
> 
> libusb support is part of STR #1575, which hasn't been merged to
> trunk yet:
> 
>     http://www.cups.org/str.php?L1575
> 

Will it be merged for CUPS 1.4?

    Till

Re: [Printing-architecture] Printer setup tools: Automatically choose the correct backend

[Klaus Singvogel]

Till,

Till Kamppeter wrote:
[...]
> 
> *cupsReplaceURI: "usb://HP/LaserJet(_|%20)42\d\d hp:/usb/HP_LaserJet_42\d\d"
> 
> Here we do not use model info from the "lpinfo -l -v" output. The 
> printer setup tool should match the first regexp against the user-chosen 
> URI and the search the "lpinfo -v" output for an URI matching the second 
> regexp.

This is not regex, it's perl extended regex you're using here. Maybe
we run into trouble, if we extend CUPS tools by perl too.

We might want to use POSIX regex please, as defined in regex(7)?

I think, the rest of the discussion has to be answered by the author
of "lpinfo" (= CUPS team). It's fine with me, if we extend CUPS in
such a way.

Regards,
	Klaus.
-- 
Klaus Singvogel  -  Maxfeldstr. 5  -  90409 Nuernberg  -  Germany
Phone: +49-911-74053-0
GnuPG-Key-ID: 1024R/5068792D  1994-06-27
SUSE LINUX Products GmbH,  GF: Markus Rex,  HRB 16746 (AG Nuernberg)

Re: [Printing-architecture] Printer setup tools: Automatically choose the correct backend

[Till Kamppeter]

Klaus Singvogel wrote:
> Till,
> 
> Till Kamppeter wrote:
> [...]
>> *cupsReplaceURI: "usb://HP/LaserJet(_|%20)42\d\d hp:/usb/HP_LaserJet_42\d\d"
>>
>> Here we do not use model info from the "lpinfo -l -v" output. The 
>> printer setup tool should match the first regexp against the user-chosen 
>> URI and the search the "lpinfo -v" output for an URI matching the second 
>> regexp.
> 
> This is not regex, it's perl extended regex you're using here. Maybe
> we run into trouble, if we extend CUPS tools by perl too.
> 
> We might want to use POSIX regex please, as defined in regex(7)?
>

Sorry for giving a bad example. All this was only a suggestion and not a 
spec. But I hope it shows the idea.

> I think, the rest of the discussion has to be answered by the author
> of "lpinfo" (= CUPS team). It's fine with me, if we extend CUPS in
> such a way.

They answered already in for them there should be only one backend per 
printer. They are working with HP now on making the "hp" backend 
unnecessary.

    Till

Re: [Printing-architecture] Printer setup tools: Automatically choose the correct backend

[Klaus Singvogel]

Till Kamppeter wrote:
> Klaus Singvogel wrote:
> >
> >We might want to use POSIX regex please, as defined in regex(7)?
> >
>
> Sorry for giving a bad example. All this was only a suggestion and
> not a
> spec. But I hope it shows the idea.

I see.
Till, thanks for clarification.

Regards,
	Klaus.
-- 
Klaus Singvogel  -  Maxfeldstr. 5  -  90409 Nuernberg  -  Germany
Phone: +49-911-74053-0
GnuPG-Key-ID: 1024R/5068792D  1994-06-27
SUSE LINUX Products GmbH,  GF: Markus Rex,  HRB 16746 (AG Nuernberg)